Dominic Olango Cagadas, Dwi Sudarno Putra, Kristine Mae Paboreal Dunque, Meri Azmi
The local coffee farmers employ manual inspection to identify the maturity of coffee cherries that are inefficient in labor and time. Thus, the objective of this study is to develop a CNN-VGG19 algorithm model that can accurately detect the maturity image of coffee cherry samples, and classify them into: unripe, semi-ripe, ripe, and overripe categories. The proposed solution will provide local coffee farmers with an automated and more accurate classification of the quality of coffee cherries. The visual geometry group-19 was employed to increase the object recognition model performance of the proposed algorithm while maintaining higher accuracy and quicker throughput, thus increasing revenues. The images are utilized as training and test set data. They were then processed by using the feature extraction of CNN-VGG19 deep learning model, and got four coffee cherry maturity classes. The model architecture attained a 90.00 % accuracy. Furthermore, the increase in both the validation and training accuracy graph with a corresponding decrease in both the validation and training loss graph propounds that the model performance has improved. © 2024, Universitas Negeri Padang. All rights reserved.
Department of Electronics Technology, University of Science and Technology of Southern Philippines, Philippines; Department of Automotive Engineering, Faculty of Engineering, Universitas Negeri Padang, Indonesia; Department of Information Technology, Politeknik Negeri Padang, Indonesia
